INDEX

Explanations

the presence of the word "Written" at the beginning of texts or articles

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ĭ±

-0.85

nel

-0.80

 Shinra

-0.78

agara

-0.71

nels

-0.70

ĪĴ

-0.70

Sensor

-0.69

allows

-0.69

illon

-0.67

alon

-0.67

POSITIVE LOGITS

escription

0.83

 written

0.76

 aloud

0.76

itatively

0.72

acters

0.72

 eloqu

0.70

 instrument

0.69

written

0.69

 intention

0.69

 tongue

0.69

Activations Density 0.027%

the presence of the word "Written" at the beginning of texts or articles

No Comments

No Known Activations